Fetch

Inspiration

We were inspired by the idea that student success leaves patterns - in grades, learning styles, and resource usage. Often, these patterns are hidden inside messy datasets or siloed systems. With Neo4j’s ability to model relationships as nodes and edges, and our love for building engaging, human-friendly UIs, we wanted to create a tool that could “fetch” these insights and bring them right to the surface.
And, of course, our pixel golden retriever mascot added a playful spin to the theme 🐕.

What it does

Fetch is a graph-powered web app that:

Finds peers with similar learning styles and shows which ones succeeded in a course.
Visualizes grade distributions, textbook usage, and learner types with interactive charts.
Uses an ML model to analyse historic data, such as courses taken by high-achieving students, and helping the student plan their trajectory.
Generates AI-powered summaries and recommendations to help students, faculty, or administrators make sense of the data.
Wraps it all in a clean UI with a pastel theme and a golden retriever mascot walking across the screen.

How we built it

Frontend: Next.js + React with Tailwind CSS and shadcn/ui for a consistent design system. Recharts for visualizations.
Backend: FastAPI serving peer queries, learner-type aggregations, KNN model for learning paths and AI summaries.
Database: Neo4j for storing and querying student-course graphs. We wrote Cypher queries to fetch similar peers based on learning-style similarity edges.
AI integration: A lightweight endpoint that consumes computed distributions and peer data, returning natural-language recommendations.

Challenges we ran into

Graph data modeling: Deciding how to represent students, courses, grades, and learning styles as nodes and relationships took iteration.
Visualization balance: At first, tables dominated the UI. We restructured around a tabbed layout so graphs and insights shine.
Time constraints: Integrating Neo4j queries, AI summaries while training the ML model and polishing the UI was a race against the clock.

Accomplishments that we're proud of

Building a full-stack graph analytics tool from scratch in a short timeframe.
Getting Neo4j queries + frontend charts + ML-powered recommendations + AI summaries all working together seamlessly.
Creating a UI that’s both professional and playful, with the retriever mascot tying the brand together.
Learning how to balance technical depth (graph algorithms, Cypher) with user experience design.

What we learned

How to design and query a knowledge graph effectively with Neo4j.
The importance of visual design choices (colors, layout, emphasis) in making insights digestible.
How to integrate multiple layers — database, API, AI/ML, and frontend — into a coherent product.

What's next for Fetch

More graph algorithms: Incorporating shortest paths, centrality, and community detection for richer peer analysis.
Recommendation engine: Personalized resource and study group suggestions powered by graph embeddings.
Scalability: Extending to larger, real-world academic datasets.
Admin dashboards: Helping universities spot course bottlenecks and design interventions.
Brand polish: Turning Fetch into a portfolio-ready, demo-friendly product with a sharper identity.

In short: Fetch started as a hackathon project, but we see it growing into a powerful knowledge graph–driven insight platform for education.